v1.3.5

• Relax spaCy constraint to <4.1.0 to support spaCy 4.
• Fix a compilation issue with recent Cython 3 versions (#415).

v1.3.4

• Extend support for transformers to v4.36.

v1.3.3

• Drop support for Python 3.6.
• Extend support for transformers to v4.35.

v1.3.2

• Extend support for transformers to v4.34.

v1.3.1

This release addresses incompatibilities related to saving and loading models across a range of transformers versions following changes in transformers v4.31.

• Extend support to transformers v4.33 (#402).
• Add fallback to load state_dict with strict=False, due to incompatibilities related to state_dict keys between transformers v4.30 and v4.31 (#398).

Options for improved pipeline package compatibility

If you are packaging models with spacy package and you'd like to avoid having your users run into the fallback strict=False loading and warnings related to state_dict keys, you may want to add a stricter transformers requirement that corresponds to the version used during training.

If you're training with the newest supported transformers version (v4.33.x), you could currently add transformers>=4.31 to your package requirements. Note that this would restrict your users to Python 3.8+ and PyTorch 1.9+ due to transformers requirements.

Create meta.json with your additional requirements:

{
  "requirements": ["transformers>=4.31"]
}

Package with spacy package:

spacy package --meta meta.json input_dir output_dir

The requirement transformers>=4.31 will be added to any other package requirements that are automatically determined based on your pipeline config.

v1.3.0

NOTE: This release has been yanked on PyPI because of the incompatibilities between saved pipelines for transformers v4.30 and v4.31. Please use spacy-transformers v1.2.x instead.

• Update build constraints to use numpy v1.25+ for python 3.9+ (#394).

• Extend support for transformers up to v4.31 (#395).

  ⚠️ After upgrading to transformers v4.31+ you may not be able to load certain older saved pipelines. All older saved models can still be loaded with transformers<4.31 or you can retrain your models so they are compatible with newer versions of transformers (#395).

  If you have created packages with spacy package using spacy-transformers v1.2 or earlier, those package requirements should be restricted to compatible versions of transformers and should not require updates.

v1.2.5

• Extend support for transformers up to v4.30.x.

v1.2.4

• Extend support for transformers up to v4.29.x.

v1.2.3

• Extend support for transformers up to v4.28.x.
• Implement coalesced pooling over entire batches (#368).

v1.2.2

• Transformer.predict: do not broadcast to listeners, requires spacy>=3.5.0 (#345)
• Correct and clarify the handling of empty/zero-length Docs during training and inference (#365)
• Remove superfluous datatype and device conversions, requires torch>=1.8.0 (#369)
• Fix memory leak in offsets mapping alignment for fast tokenizers (#373)